has secured substantial growth capital at a valuation now 13 times greater than its previous funding round, reflecting confidence that the next frontier in artificial intelligence will hinge less on enhanced language models and more on exclusive data assets combined with causal reasoning capabilities.
Based in San Francisco, this innovative startup specializes in AI systems designed to uncover causation rather than mere correlation. A significant portion of the new investment is earmarked for deploying what it claims to be one of the fastest privately owned supercomputers globally-an advanced liquid-cooled infrastructure that will underpin its enterprise-grade causal AI solutions.
The funding round, spearheaded by leading venture firms with participation from prominent investors, places Alembic among an elite cadre of AI labs revolutionizing how large organizations make high-stakes decisions worth millions.
From Modest Beginnings to a Paradigm Shift in AI
Since its inception, Alembic has undergone a remarkable transformation, reflected in its valuation surge from approximately $50 million to around $645 million. Initially, the company focused on signal processing and correlation analytics within marketing measurement. Causal inference was not part of its toolkit until after its Series A round.
Founder and CEO Tomás Puig recalls the early days when the startup was so financially constrained it couldn’t even run simulations to validate its causal models. Testing began on a makeshift cluster of Mac Pros, lacking GPU acceleration. To their surprise, the causal models proved effective not only in marketing but across diverse business domains involving time-series data.
Iterative development driven by customer feedback revealed that what was initially conceived as a vertical-specific solution evolved into a generalized foundational model applicable enterprise-wide-spanning sales, marketing, supply chain, finance, and beyond.
The Critical Role of Causal AI in Enterprise Decision-Making
Unlike traditional analytics tools that identify correlations-such as linking social media engagement with sales spikes-Alembic’s causal AI distinguishes true cause-and-effect relationships. For example, it can determine whether a viral news event or social media campaign actually drove sales increases, a distinction vital for strategic budget allocation.
Puig emphasizes that while companies are awash in data, actionable insights remain scarce. Alembic’s platform now serves a diverse clientele, including Fortune 500 firms across financial services, technology, and consumer packaged goods, enabling them to answer complex questions about marketing ROI, operational efficiency, and investment strategies.
Delta Air Lines’ Chief Marketing Officer Alicia Tillman highlights the platform’s transformative impact: “Alembic connected marketing exposure directly to business outcomes with unprecedented speed and precision, providing a unified view across channels that traditional tools couldn’t deliver.” The airline quantified revenue uplift from brand campaigns within days, a feat long elusive in marketing measurement.
Engineering a Supercomputer Tailored for Causal AI
To meet the demanding computational needs of its causal models and stringent data privacy requirements, Alembic opted to build its own liquid-cooled supercomputer, equipped with Nvidia’s cutting-edge Blackwell GPUs. This system, housed in a neutral data center in San Jose, California, is reportedly unique among non-Fortune 500 companies.
Unlike large language models trained once on static datasets, Alembic’s AI employs “online and evolving” spiking neural networks-brain-inspired architectures that continuously adapt as new data streams in. This dynamic learning process generates a distinct “brain” for each client, enabling highly customized insights.
The system explores billions of data permutations to identify the strongest causal signals, demanding “F1 car” level infrastructure rather than conventional cloud solutions. Custom CUDA kernels and low-level optimizations tailored for causal inference workloads further enhance performance.
Such intensive usage once caused GPUs to overheat and fail, prompting Nvidia to fast-track liquid-cooled hardware access. Owning this infrastructure also offers significant cost advantages; Alembic estimates that equivalent cloud compute would cost upwards of $62 million annually, whereas their owned setup is a fraction of that expense.
Moreover, operating proprietary infrastructure addresses critical data sovereignty concerns. Many clients in regulated sectors, including financial services and consumer packaged goods, prohibit storing sensitive data on public cloud platforms like AWS, Microsoft Azure, or Google Cloud. Alembic’s neutral data center approach thus creates a competitive moat difficult for hyperscale cloud providers to overcome.
The Pivotal Nvidia Partnership That Accelerated Growth
Alembic’s collaboration with Nvidia began unexpectedly after its Series A announcement, when Nvidia’s CEO Jensen Huang encouraged his team to explore the startup. Initial contact was made via LinkedIn due to the absence of a formal inquiry channel on Alembic’s website.
Early computational limitations nearly stalled progress; generating causal analyses took weeks on Mac Pros, making regular reporting infeasible. Nvidia intervened by facilitating access to a private data center cage and providing the first GPU cluster, a move Puig credits with enabling the company’s survival and growth.
Today, Alembic leverages Nvidia’s comprehensive software stack, including cuGraph for graph analytics and TensorRT for accelerated inference. This deep integration allows Alembic’s researchers to focus on pioneering causal AI methodologies rather than low-level engineering challenges.
When Alembic’s intense workloads caused repeated GPU failures, Nvidia responded by expediting access to next-generation liquid-cooled systems, underscoring the strength of their partnership.
Expanding Impact: From Viral Candy Campaigns to Financial Services
Alembic’s client base has rapidly grown to include major corporations across sectors such as technology, finance, consumer goods, and even collegiate athletics. Use cases have diversified far beyond marketing analytics.
Mars Incorporated utilized Alembic’s platform to measure the sales impact of altering candy shapes during promotional events. A Fortune 500 tech firm boosted its sales pipeline by 37% through Alembic’s attribution models. Financial institutions have linked CEO public appearances and co-marketing efforts directly to fund inflows, providing unprecedented clarity on marketing effectiveness.
Gülen Bengi, Global Chief Marketing Officer at Mars Snacking, describes the technology as “liberating creativity through mathematics,” enabling the company to quantify the sales impact of viral organic conversations that were previously only qualitatively understood.
Alembic’s platform can forecast revenue, customer acquisition, and deal closures up to two years ahead with 95% confidence, transforming how enterprises evaluate investments like stadium naming rights or major sponsorships-decisions once made without concrete financial attribution.
Why Industry Giants Face Challenges Replicating Alembic’s Success
Operating in a competitive landscape that includes established marketing analytics providers and emerging AI startups, Alembic maintains several structural advantages. Its proprietary causal inference mathematics, protected by patents, require years of specialized research to replicate.
The company’s enormous computational demands create a natural barrier to entry, with cloud-equivalent costs estimated in the hundreds of millions annually. Additionally, Alembic’s commitment to data sovereignty through neutral infrastructure appeals to clients wary of entrusting sensitive data to hyperscale cloud vendors, a concern growing among enterprises.
Years of engineering to handle fragmented, messy enterprise data preceded Alembic’s causal AI breakthroughs, providing a foundation that competitors would find difficult to duplicate quickly.
Reimagining Enterprise AI: The Power of Private Data and Causal Insight
The recent $145 million funding round underscores a contrarian vision in an AI ecosystem largely focused on scaling language models. While companies like OpenAI and Google compete to build more capable chatbots, Alembic is constructing infrastructure that deciphers cause-and-effect relationships within proprietary corporate data-information that generic models cannot access.
Alembic’s journey from a cash-strapped startup running simulations on Mac Pros to operating one of the world’s fastest private supercomputers exemplifies the maturation of enterprise AI. As AI transitions from experimental to mission-critical, organizations demand systems tailored to their unique data environments rather than generic public models.
Puig draws parallels to other technological evolutions: just as search engines commoditized public information, elevating the value of proprietary data, and cloud computing transformed infrastructure into a utility, the convergence of large language models shifts competitive advantage to those who can extract unique intelligence from private datasets.
Beyond marketing, Alembic is piloting applications in robotics, where causal models could enable autonomous systems to understand the consequences of their actions. New offerings, such as GPU-accelerated databases, are also emerging, with the ambition to become the “central nervous system” connecting cause and effect across all enterprise functions.
While challenges remain-including lengthy sales cycles, integration complexities, and evolving competition-the company’s unique position is reinforced by marquee clients, proprietary technology, and substantial capital to scale its vision of shifting enterprise AI from correlation to causation.
Puig likens Alembic’s approach to quantitative hedge funds like Renaissance Technologies, which leverage sophisticated mathematical models to achieve returns beyond the reach of general-purpose AI. “ChatGPT still can’t match Renaissance Technologies,” he notes, highlighting the enduring value of specialized intelligence.
This analogy encapsulates Alembic’s core insight: in a world where everyone accesses similar general AI tools, sustainable advantage arises from systems that decode the hidden cause-and-effect dynamics within proprietary data-ensuring competitors cannot simply replicate answers to critical business questions.
